Is Claude Free? Claude Pricing (2026): Where the Free Plan Stops

Yes. Claude Free costs $0 and is capable enough for occasional research, writing, file work, and web search, but its usage resets on a rolling five-hour window and Anthropic publishes no fixed daily message count. Most people should stay free until a reset interrupts valuable work; Claude Pro starts at $20 monthly or $200 paid up front for a year.

Claude's prices and limits below were verified against Anthropic's live Claude pricing page and current help pages on 2 August 2026. Prices are current US list prices before applicable tax.

Claude pricing at a glance

Free is the right default, Pro is the practical work plan, and Max is a capacity purchase. Team and Enterprise solve different problems: Team sells governed seats with included usage, while Enterprise charges a seat fee and meters every token separately.

The table is a price map, not a reason to upgrade. The decision turns on one observation: does the current limit stop an outcome worth more than the price difference? If the answer is no, paying buys unused capacity.

What Claude Free actually includes

Claude Free is a continuing product, not a countdown to a paid trial. At $0, Anthropic currently includes chat on the web, iOS, Android, and desktop, plus writing, code generation, data visualization, web search, memory across conversations, file creation, code execution, desktop extensions, Slack and Google Workspace connections, remote MCP connectors, and extended thinking.

That is enough for a senior operator who asks a few research questions, rewrites a document, checks a spreadsheet, or summarizes a meeting packet without keeping Claude open all day. It is also enough for a founder evaluating whether Claude's working style fits before assigning it a recurring place in the budget.

The limit is capacity, not a missing timer. Anthropic says every plan resets usage on a rolling five-hour session window. Consumption changes with conversation length, attached files, the model, and the feature being used. A long conversation that repeatedly carries a large document costs more capacity than a sequence of short questions.

Free stops being comfortable when work must continue after a cap, when priority access matters at a busy time, or when you need paid surfaces such as Claude Code, Claude Cowork, Claude Design, Claude Science, unlimited projects, Research, more models, or Claude for Microsoft 365. Those are Pro additions on the current price page.

Who never needs to pay: anyone whose useful work fits inside Free and who can wait through a reset without losing money, a deadline, or concentration. A casual user is not missing a secret value tier. Free remains the rational choice until the limit becomes measurable friction.

What each paid Claude plan buys

Pay for the smallest plan that removes a demonstrated constraint. Each step changes either capacity, work surfaces, or governance. Confusing those three produces the expensive upgrades.

Claude Pro: the sensible work plan

Claude Pro earns its $20 monthly price when Free interrupts useful work. It provides at least five times Free usage per five-hour session and adds weekly limits rather than unlimited access. The Pro help page also warns that Anthropic may apply model, feature, weekly, or monthly caps to manage capacity.

Pro adds Claude Code for terminal work, Cowork for delegated desktop tasks, Claude Design, Claude Science, unlimited projects, Research, more model access, and Claude for Microsoft 365. A solo technical builder who uses chat and Claude Code in the same day should notice one important constraint: activity across Claude's app surfaces and Claude Code draws from the same usage pool.

The plan is $20 month to month or $200 paid up front for a year. The annual option lowers the arithmetic price to $16.67 per month, although Anthropic rounds the display to $17. The discount is worthwhile for steady use, not for a short project.

Pro does not include Claude API usage. A founder using Pro for interactive work and an application using the Claude API can receive two separate bills.

Claude Max: more flow, not unlimited Claude

Claude Max is justified only after Pro exhaustion is frequent. Max 5x costs $100 per month and carries five times Pro usage per session. Max 20x costs $200 per month and carries twenty times Pro usage per session.

Both tiers add higher output limits, priority during busy periods, and early feature access. Both still have a weekly limit across all models and a second weekly limit for Sonnet models. Anthropic can also apply other model, feature, weekly, or monthly caps. The word "Max" is not an unlimited-use promise.

Max is monthly only, and mobile subscription prices can vary by platform. That flexibility is useful for a funded founder facing a temporary launch month: move up for a known heavy period, then move down. It is poor value for someone buying reassurance against a cap they rarely hit.

Claude Team: pay for control and per-member capacity

Claude Team is the right purchase when identity, billing, connectors, and administration matter alongside usage. The plan requires at least 2 members and supports up to 150 seats.

Standard seats cost $25 per member monthly or $20 per member per month on annual billing. Each Standard member receives 1.25 times Pro usage per session and an individual weekly limit across all models.

Premium seats cost $125 monthly or $100 per member per month on annual billing. Each Premium member receives 6.25 times Pro usage per session, plus weekly all-model and Sonnet limits. Team usage is per person, so one heavy user does not drain another person's included allowance.

The governance layer includes central billing, SSO, connector controls, workplace connectors, and enterprise search. A mid-market CTO should buy Team because those controls reduce operational risk, not because several separate Pro accounts look untidy. Standard and Premium seats can be mixed so the expensive capacity goes only to people who use it.

Claude Enterprise: the seat is only the cover charge

Claude Enterprise is metered from the first token. The current public structure is $20 per seat per month, billed annually, plus all usage at standard API rates. The Enterprise help page is explicit: the seat includes access, not a token allowance.

The current usage-based plan has no plan-level or seat-level usage cap. Admins control exposure with organization and user spend limits instead. Self-serve Enterprise requires at least 20 seats; sales-assisted Enterprise requires at least 50.

At the self-serve minimum, seat access alone is $400 per month or $4,800 for the annual commitment, before one token is used. That calculation is why Enterprise belongs in a procurement decision, not an individual price comparison. Its case is advanced security, compliance, administration, a published 500k context window on the default model, and metered scale.

The upgrade break-even

Claude Pro's annual plan becomes cheaper only when you keep it for more than 10 paid months. Ten monthly payments equal the $200 annual price. Month 11 takes monthly spending to $220; a full 12 months costs $240, so annual billing saves $40, or 16.67%.

Price per outcome makes the subscription easier to judge. If Pro supports 100 completed outcomes in a month, its $20 price allocates to $0.20 each. If it supports only 20, the allocation is $1 each. An outcome means a finished analysis, approved document, solved code task, or decision, not another prompt.

Max exposes a different calculation. Max 5x charges $100 for five Pro-capacity multiples, or $20 per multiple. Max 20x charges $200 for twenty multiples, or $10 per multiple. The $200 plan has the better unit price, but it is cheaper only in the abstract if most of the allowance goes unused.

Team annual billing has a cleaner break-even. Standard costs $240 per seat-year instead of $300 across 12 monthly payments, saving $60 or 20%. Premium costs $1,200 instead of $1,500, saving $300 or 20%. Both annual options become cheaper in month 10, so annualize stable seats rather than planned hires.

The decision rule is blunt. A senior operator whose Free limit interrupts one $100 analysis can justify Pro immediately. A solo builder who reaches Pro limits twice a quarter should buy usage credits for those bursts before locking in Max. A team that needs SSO should evaluate Team even when individual Pro capacity is sufficient, because the purchase is governance rather than messages.

Hidden Claude costs and billing catches

The sticker price stops being the whole bill once usage credits, annual commitments, taxes, and purchase channels enter the picture.

Extra usage can turn a fixed plan into metered spend

Pro and Max subscribers can enable usage credits after reaching included limits. Continued work is then billed at standard API rates and appears as a separate charge. You prepay the balance, and the normal five-hour reset still happens.

Anthropic provides spend alerts, an auto-reload threshold, and a monthly spending cap. It also lets you set that cap to unlimited. That last option removes the safety rail, so a paid subscription can quietly become a subscription plus open-ended token spend. Research, large project files, long conversations, and Claude Code can consume credits faster because they carry more context. Individual redemptions are limited to $2,000 per day.

Prepaid usage bundles reduce the rate: $50 of value costs $45, $250 costs $200, and $1,000 costs $700. Those are 10%, 20%, and 30% discounts. Pro and Max can buy up to $2,000 of discounted bundles per month; Team can buy up to $3,000. Usage beyond those caps returns to standard rates.

Do not confuse app usage balances with Claude Console API credits. Anthropic's API credit page says Console credits expire one year after purchase, cannot be extended, and are non-refundable. Buying a large speculative API balance can convert a volume discount into expired cash.

Annual discounts exchange price for flexibility

Pro's $40 annual saving is earned only if the subscription survives the year. Team's 20% discount has the same condition at every seat. An annual seat that no longer has an owner can erase the saving faster than the lower monthly rate creates it.

Max offers no annual discount, which is a benefit for uncertain heavy use. It also means a full year of Max 5x costs $1,200 and Max 20x costs $2,400. Those totals are not discounted commitments; they are twelve monthly decisions.

Taxes, mobile pricing, and refunds change the checkout number

Anthropic's US prices exclude applicable tax. Local currency, tax handling, and mobile pricing can vary by region and platform, so the web price is the clean comparison baseline rather than a guarantee for every checkout.

Payments are generally non-refundable unless Anthropic's terms or local law require otherwise. Customers in the EEA and UK may receive a usage-prorated refund within 14 days. Apple handles subscriptions purchased through iOS, while Android refund handling depends on purchase status and Google support rules.

The safe practice is simple: buy annual only for stable demand, keep auto-reload capped, purchase API credit near expected use, and subscribe on the channel whose refund path you understand.

Claude app pricing versus API pricing

Claude subscriptions buy an interactive product; the API buys metered model calls for software. Pro and Max do not include API tokens, and API credits do not add Pro or Max app capacity.

Anthropic's current model pricing lists Fable 5 at $10 per million input tokens and $50 per million output tokens. Opus 5 is $5 and $25. Haiku 4.5 is $1 and $5.

Sonnet 5 carries time-sensitive pricing. Through 31 August 2026 it costs $2 per million input tokens and $10 per million output tokens. On 1 September, those rates become $3 and $15.

For a normalized task with 10,000 input tokens and 2,000 output tokens, Sonnet 5 costs $0.04 through August and $0.06 from September, a 50% increase. The same request shape costs $0.10 on Opus 5, $0.02 on Haiku 4.5, and $0.20 on Fable 5. These are price comparisons for one request shape, not claims that the models deliver equal quality.

At the introductory Sonnet 5 rate, $20 buys 500 of those normalized tasks. From September it buys about 333. That makes the API attractive for intermittent automation, while Pro remains the cleaner purchase for an interactive workspace with projects, Research, Claude Code, and other paid surfaces.

There is a second Sonnet 5 cost change hidden behind the rate card. Anthropic says Claude 4.7 and later use a tokenizer that produces approximately 30% more tokens for the same text, with the exact effect varying by workload. If that 30% estimate holds at the same input and output mix, Sonnet 5 is about 13.3% cheaper than Sonnet 4.6 for the same text during the introductory period. From September, matching $3/$15 rates would make that same text roughly 30% more expensive before discounts.

API modifiers can materially change the result. Five-minute cache writes cost 1.25 times base input, one-hour cache writes cost 2 times, and cache reads cost 0.1 times. Batch processing cuts input and output token charges by 50%. US-only inference on Claude 4.6 and later costs 1.1 times standard pricing, while partner regional or multi-region endpoints carry a 10% premium over global endpoints. Web search adds $10 per 1,000 searches on top of token costs.

For terminal-specific usage patterns, the Claude Code pricing guide goes deeper on where coding sessions consume the allowance.

Claude pricing versus ChatGPT and Gemini

Claude is cheapest at the mainstream paid tier only when you accept its annual commitment. Monthly Claude Pro and ChatGPT Plus both cost $20; Google AI Pro costs $19.99.

ChatGPT's current personal ladder is $0 for Free, $8 for Go, $20 for Plus, and $100 for Pro. A buyer who needs a paid option below $20 has Go as a cheaper entry point, while Claude's paid individual ladder jumps directly from Free to Pro.

Claude Pro's $200 annual commitment normalizes to $16.67 per month. That is $3.33 below ChatGPT Plus, or $40 across a full year. Choose Claude on that price advantage only if its work surfaces fit and 11 to 12 months of use are likely. Choose ChatGPT Plus when a monthly commitment or its product set is worth the difference. The broader product decision is covered in Claude vs ChatGPT.

Google Gemini's personal ladder starts at $4.99 for Google AI Plus, then $19.99 for Google AI Pro. Google AI Ultra 5x is $99.99 and Ultra 20x is $199.99.

Against Google AI Pro, annual Claude Pro is $3.32 lower per month, or $39.88 per year. Without the annual commitment, Claude costs one cent more. At the heavy-use tier, Claude Max 5x equals ChatGPT Pro at $100 and costs one cent more than Google AI Ultra 5x. The price bands are effectively tied; workflow fit decides the winner.

The normalized call is clear:

• Choose Claude Free while its resets stay outside important work.
• Choose Claude Pro annual for the lowest normalized mainstream price when Claude is a year-round tool.
• Choose ChatGPT Go or Google AI Plus when the budget ceiling is below $20 and their respective product surfaces fit.
• Choose among the roughly $100 heavy-use tiers only after measuring exhaustion on the mainstream plan.

For a wider product shortlist, see the best AI chatbot comparison.

Which Claude plan should you choose?

Use the smallest Claude plan that clears a named constraint.

• Stay on Free when resets are tolerable, the free capabilities cover the job, and waiting does not damage an outcome.
• Choose Pro monthly for uncertain or seasonal professional use lasting 1 to 9 months.
• Choose Pro annual for 11 to 12 expected months of use; 10 months is a tie.
• Choose Max 5x after Pro limits repeatedly interrupt high-value individual work and usage credits no longer suit the pattern.
• Choose Max 20x only when usage approaches the 20x allowance. Its unit economics are strong, but idle capacity is still waste.
• Choose Team when at least 2 people need administration, SSO, workplace connectors, and individual governed capacity.
• Choose Enterprise when at least 20 seats need advanced controls and metered usage is preferable to included caps.

The final check is economic, not aspirational. If Free produces the outcomes you need, it is the best Claude plan. If an interruption costs more than $20, Pro becomes easy to defend. Max, Team, and Enterprise need measured usage or governance requirements, not curiosity.

Frequently asked questions

Is Claude AI still free?

Yes. Claude Free is a continuing $0 plan available in supported locations. It has variable usage limits, but it is not presented as a timed free trial.

How long can you use Claude for free?

The Free plan has no published expiration date. Included usage refreshes on a rolling five-hour window, so you can continue after capacity resets.

Does Claude have a daily usage limit?

Anthropic does not publish a fixed daily message number. Usage varies with conversation length, files, models, features, traffic, and account conditions; paid plans also add weekly limits.

Is Claude free for personal use?

Yes. Individuals can use Claude Free for $0. Pay only when the free capacity or feature set stops work that is worth more than the upgrade.

Is Claude free for students?

Students can use the standard Free plan. Anthropic lists no standing individual discount for Pro or Max, while the Education offer is a university-wide institutional plan for students, faculty, and staff.

Can I get a refund for Claude Pro or Max?

Payments are generally non-refundable except where terms or local law require otherwise. EEA and UK customers may receive a usage-prorated refund within 14 days, and Apple handles iOS purchases directly.

Is Claude API usage included with Pro or Max?

No. Pro and Max subscriptions cover Claude's consumer app surfaces and included Claude Code access. API usage uses separate prepaid Console credits and token-based pricing.

Has Claude pricing changed in 2026?

The time-sensitive change is Claude Sonnet 5 API pricing. Its introductory $2/$10 per million input/output token rate ends on 31 August 2026; the standard $3/$15 rate begins on 1 September.

Want a faster way to match AI spend to the job each tool should do? Get the AI Tools Map for Business Owners.